A reset-type Kalman filter for the navigation of a winged space vehicle is discussed and a triple rate Kalman filter is proposed to lighten the computation load. The basic idea is to pay attention to the existence of fast and slowly varying state variables and thus to partition the covariance matrix of the conventional Kalman filter into parts of smaller dimensions which are propagated in parallel at three different (fast, middle and slow) rates. Because of the reduction in the propagation period for the fast rate part, the approximation accuracy of time-varying system can be improved for a given computer load and a better estimation accuracy is realized. These are verified by computer simulations for the case of a reusable winged vehicle. The triple-rate filter can be extended to a multi-rate version.